controlling the level of verbosity in Hive

问题:

I run hive from command line that way:

hive -e "select * from table;"

and it gives me all the details and stage of my query. If I do:

hive -S -e "select * from table;"

It doesn't give any detail.

What I would like to have is just to be shown the numbers of mappers and reducers for each step of map reduce, as well as the number of map reduce steps, but I don't want to see all the informations like:

2013-01-02 12:46:28,756 Stage-1 map = 100%,  reduce = 67%, Cumulative CPU 287.99 sec

So, is it possible to have a more granular control of the verbosity with Hive? Thanks!

回答1:

My "solution" to the same problem is

hive ... 2>&1 | grep --line-buffered -v -e ' map = .*,  reduce = ' | tee log

This sucks (and might kill more than necessary), but this is the best I have been able to come up with so far.

EDIT: please watch and vote for The logging (progress reporting) is too verbose